Methyltransferase Set7/9 Maintains Transcription and Euchromatin Structure at Islet-Enriched Genes

Polyclonal antibody against monomethyl–H3 Lys4 was from Abcam, and antibodies against dimethylated H3 Lys4, Pdx1, and Set7/9 were from Millipore; anti-RNA polymerase II COOH-terminal domain was from Covance; anti-actin, anti-glucagon, and anti-insulin were from Santa Cruz Biotechnology. βTC3 cells, INS-1 (832/13), αTC1.6, and NIH3T3 were maintained as previously described (20,21). Islets were isolated from C57BL/6J mice following collagenase digestion of pancreatic tissue as previously described (22). Techniques and protocols were approved by the Indiana University Institutional Animal Care and Use Committee.

Real time RT-PCR was performed using forward and reverse primers to amplify insulin premRNA (5′-GGGGAGCGTGGCTTCTTCTA-3′ and 5′ GGGGACAGAATTCAGTGGCA-3′) or Actb (5′-AGGTCATCACTATTGGCAACGA-3′ and 5′-CACTTCATGATGGAATTGAATGTAGTT-3′) as previously described in detail (23). Primers for Setd7, Slc2a2, Gck, Nkx6–1, NeuroD1, Foxo1, Pax6, Iapp, Mafa, Ipf1, Kcnj11, Abcc8, and Foxa2 were obtained from Qiagen (Quantitech primer kit catalog nos. QT00139727, QT00103537, QT00140007, QT00143318, QT00251265, QT00116186, QT01052786, QT00101052, QT01037638, QT00102235, QT00305319, QT01042300, and QT00242809, respectively). PCR products were subcloned and sequenced to verify amplification of the correct gene product. All values are normalized to Actb mRNA levels.

Whole-cell extracts were prepared from homogenized tissues, 50 islets, or 1 × 10⁶ βTC3 cells by lysis in a buffer containing sodium dodecyl sulfate, and 5 μg extract was resolved by electrophoresis on a 12% SDS-polyacrylamide gel followed by immunoblot with anti-Set7/9 or anti-actin antisera. Immunoblots were visualized using the ECL-Plus system (Amersham).

Chromatin immunoprecipitation (ChIP) assays were performed as previously described (20). Briefly, 10⁷ βTC3 cells or 300 mouse islets were fixed in 1% formaldehyde to cross-link proteins to DNA, and cells were lysed and chromatin sheared to 400–600 base pairs (bp) by sonication. Antibodies were used to subsequently immunoprecipitate the protein of interest and then recovery of coimmunoprecipitated DNA fragments was assessed using SYBR green–based quantitative real-time PCR. Data are expressed as the percent recovery of coimmunoprecipitated DNA relative to input DNA (before immunoprecipitation). Primers used for amplification were as follows: for the Ins1/2 promoter, 5′-TACCTTGCTGCCTGAGTTCTGC-3′ forward and 5′-GCATTTTCCACATCATTCCCC-3′ reverse; the Ins1/2 coding region, 5′-TGGCTTCTTCTACACACCCAAG-3′ forward and 5′-ACAATGCCACGCTTCTGCC-3′ reverse; the Slc2a2 promoter, 5′-ATCTGGCTCCGCACTCTCATCTTG-3′ forward and 5′- CCCTGTGACTTTTCTGTGTCTTAGG-3′ reverse; the Pdx1 promoter, 5′-AGCCTGGACTACATCTTGAGTTGC-3′ forward and 5′-TTACAAAACCGCTCCCAATGAG-3′ reverse; and the Setd7 promoter, 5′- TCCTGTTGCCATCATCAAAA-3′ forward and 5′-ACATGAAGGCACCACATGAA-3′ reverse.

Mouse pancreata were fixed by cardiac perfusion with 4% paraformaldehyde, paraffin embedded, and sectioned at 5-μm intervals. Human pancreatic samples were obtained from the University of Virginia Human Islet Transplant Program, fixed for 2 h in 4% paraformaldehyde, and processed similarly to mouse pancreata. Immunohistochemical analysis of insulin (anti-insulin, 1:500 dilution), Pdx1 (anti-Pdx1, 1:2,000 dilution), glucagon (anti-glucagon, 1:500 dilution), and Set7/9 (anti-Set7/9, 1:50), was performed as previously described (11). For immunofluorescence, secondary antibodies were goat anti-rabbit–conjugated to Alexa Fluor 555 and donkey anti-mouse–conjugated to Alexa Fluor 488 (Molecular Probes). Images were acquired using a Zeiss Z1 microscope equipped with Apotome optical sectioning hardware.

siRNAs against Set7/9 or a scrambled control were mixed with Reagent V (Amaxa, Inc.) and transfected into 3 × 10⁶ βTC3 mouse insulinoma cells using an Amaxa Nucleofector (program d-23) according to the manufacturer's protocol. Cells were then plated in a six-well dish containing culture medium for 72 h and then harvested for protein, total RNA, or ChIP analysis. siRNA sequences used were as follows: siSet97, 5′-GGUUUAUGUUGCCGACUCTT-3′; siSet98, 5′-GGUAGCAGUUGGACCUAAUTT-3′; and a control siRNA (siControl), 5′-AAAGUCGACCUUCAGUAAGUU-3′.

Stabilized siRNAs were obtained from Dharmacon and diluted in 0.9% saline. Eight-week-old male C57BL/6J mice were injected intraperitoneally with 1.6 mg/kg mouse wt siRNA daily for 4 days. Islets were then harvested, handpicked and allowed to recover overnight and then subjected to glucose-stimulated insulin secretion (GSIS)/glucose-stimulated Ca²⁺ mobilization (GSCa) studies (see below), protein isolation, or RNA isolation. siRNA sequences used were as follows: siSet1, 5′-CGGACAGAUUUGAGGGCAAUU-3′; siSet2, 5′-GGGUUUAUGUUGCCGACUCUU-3; and siControl, 5′-AAAGUCGACCUUCAGUAAGUU-3′.

For GSIS studies, 50 islets per condition were incubated in Krebs-Ringer HEPES buffered solution for 1 h at 37°C and then placed in Krebs-Ringer HEPES buffered solution containing 3 or 11 mmol/l glucose for 1 h. Insulin released into the medium was assayed using a two-site immunospecific ELISA (Alpco Diagnostics). GSCa in mouse islets was measured using the ratiometric Ca²⁺ indicator Fura-2AM as described previously (24). GSCa was defined as the difference between ratio measurements (340/380 nm fluorescence) in 11 mmol/l vs. 3 mmol/l glucose. Data were analyzed with IP Lab software, version 4.0 (Scanalytics).

A genomic clone containing ∼8,000 bp of the Setd7 gene was used as a template in PCR reactions to generate various fragments of the Setd7 gene, which were then subcloned upstream of the luciferase coding sequence in plasmid pFoxLuc (25). Mutagenesis of the Setd7 gene was performed using the QuickChange Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer's instructions. All sequences were confirmed by automated DNA sequencing.

10⁶ βTC3/αTC1.6/INS-1 cells or 5 × 10⁵ NIH3T3 cells were seeded in six-well plates 24 h before transfection. Plasmid (2 μg) diluted into 0.1 ml PBS was mixed with 6 μl of Metafectene Pro (Biotex) and incubated at room temperature for 15 min. The transfection mixture was added to each well of a six-well plate along with 2 ml medium without antibiotics. At 6 h after transfection, the medium was replaced with fresh medium containing antibiotics. At 48 h after transfection, cells were harvested and a luciferase assay was performed utilizing a commercially available luminometric kit (Promega). Luciferase activity was normalized to protein concentration as measured using the Bio-Rad Protein Assay reagent (Bio-Rad) according to manufacturer's instructions.

To evaluate the tissue distribution of Set7/9 in adult mice, Set7/9 mRNA and protein were assayed in different tissues by real-time RT-PCR and immunoblot, respectively. As shown in Fig. 1A, mRNA levels demonstrate a pattern of tissue-specific enhancement, with the highest relative levels seen in muscle, brain, heart, and pancreas. Interestingly, Set7/9 protein levels do not completely mirror Set7/9 mRNA levels (Fig. 1B), as no protein was detectable in either heart or total pancreas and very little was detectable in muscle. These data suggest the potential for posttranscriptional regulation of Set7/9 expression. However, within the pancreas, there was clear enrichment of Set7/9 protein within islets as determined by immunoblotting (Fig. 1B). To directly visualize the distribution of Set7/9 among pancreatic cell types, we next performed immunohistochemistry of pancreatic sections from both mice and humans. Figure 1C and D shows that Set7/9 is strikingly enriched in islets of mouse pancreas, with a predominantly nuclear pattern in most cells of the central islet. Coimmunofluorescence revealed that Set7/9 exhibited a nuclear pattern in insulin-producing β-cells (Fig. 1E–G) and both a nuclear and cytoplasmic pattern in glucagon-producing α-cells (Fig. 1H–J). All Pdx1-positive cells of the islet costained for Set7/9 (Fig. 1K–M). Human pancreas sections demonstrated a Set7/9 staining pattern identical to that seen in mouse sections (supplemental Fig. S1, available in an online appendix at http://dx.doi.org/10.2337/db08-1150).

The islet-specific enrichment of Set7/9 suggested the potential for regulation at the transcriptional level. To explore this possibility, we placed a range of DNA fragments from the 5′ regulatory region of the gene encoding Set7/9 (Setd7) upstream of the luciferase reporter gene (in plasmid pFoxLuc [25]). These constructs were then transfected into the β-cell–derived cell lines βTC3 and INS-1, the α-cell–derived cell line αTC1.6, and the fibroblast-derived line NIH3T3. As shown in Fig. 2A, β-cell lines displayed enhancement of luciferase activity with a DNA fragment containing −6,584 bp upstream of the transcriptional start site. This enhancement was significantly attenuated when a fragment containing −3,131 bp from the transcriptional start site was used, suggesting the potential for a β-cell–specific enhancer located between −3,131 and −6,584 bp. To further explore this possibility, we next performed an alignment of human, mouse, and rat Setd7 genes to identify regions of homology that might suggest conserved β-cell–specific enhancer regions. Although the entire region between −3,131 and −6,584 bp only exhibited ∼25% identity between these species (data not shown), we identified a region between −5,768 and −6,030 bp that exhibited ∼68% identity (supplemental Fig. S2). Within this conserved region, we identified two potential binding sites for the β-cell–specific transcription factor Pdx1 (consensus: 5′-TAAT-3′ [refs. 20,26]) (supplemental Fig. S2). To directly test whether this region of the Setd7 gene might be regulated in β-cells, a DNA fragment containing this conserved region was placed upstream of the prolactin minimal promoter driving luciferase and used in reporter gene analysis studies. As shown in Fig. 2B, this fragment displayed relative enhancement of luciferase activity in βTC3 and INS-1 β-cell lines,but not in NIH3T3 cells. Upon cotransfection with a plasmid encoding Pdx1, this fragment displayed an ∼25-fold enhancement in the β-cell lines INS-1 and βTC3 (Fig. 2C and D). Mutation of either putative Pdx1 binding site attenuated activity of this fragment in both β-cell lines (Fig. 2C and D). Pdx1 appears to occupy the endogenous Setd7 gene in βTC3 cells within the region around −6,000 bp, as assessed by the ChIP assay, but does not occupy the endogenous gene encoding β-actin (Fig. 2E). Together, these results suggest that the Setd7 gene may contain elements that enhance its expression specifically in β-cells and that the islet transcription factor Pdx1 may be a necessary component of a transactivation complex on the Setd7 gene.

To determine the role of Set7/9 in the regulation of β-cell–specific genes, we next depleted βTC3 cells of Set7/9 protein and analyzed gene expression by real-time RT-PCR. To knock down protein, two different siRNAs (siSet97 and siSet98) targeted against murine Set7/9 or siControl were transfected into βTC3 cells. As shown in Fig. 3A, >90% reductions in Set7/9 protein were seen following transfections with each siRNA compared with those seen in control transfections. Use of two different siRNAs ensured that any effects on gene transcription were unlikely a result of off-target effects of these siRNAs. A survey of islet-specific genes involved in glucose responsiveness showed that the genes encoding the β-cell–specific transcription factor MafA (MafA), the glucose transporter Glut2 (Slc2a2), and preproinsulin (Ins1/2) were reduced significantly (Fig. 3B) in cells depleted of Set7/9. However, reduction was not seen uniformly among all genes responsible for glucose responsiveness in islets, as those encoding Pdx1 (Pdx1), the ATP-sensitive K⁺ channel (Kcnj11), sulfonylurea receptors (Abcc8), Nkx6.1 (Nkx6.1), and NeuroD1 (NeuroD1) appeared statistically unaffected by acute reductions in Set7/9 (Fig. 3B).

To confirm our findings in primary cells, we next performed knockdown experiments in mouse islets. However, a major challenge in RNA intereference studies in primary rodent islets is their relatively poor transfection efficiency in vitro, which typically necessitates use of viral-based approaches (27). Instead, we sought to knock down Set7/9 in islets by delivering chemically stabilized siRNAs in vivo. A similar technique has been used to target other proteins for systemic knockdown (28,29). To test the efficacy of this approach in islets, we performed four daily intraperitoneal injections of Cy3-labeled double-stranded RNA into male C57BL/6J mice and subsequently isolated islets from these animals on day 5. As shown in Fig. 4A, fluorescent microscopic images of isolated islets revealed penetration of the Cy3 label into the islets, suggesting islet entry of siRNA. Two different stabilized siRNAs targeted against Set7/9 (siSet1 and siSet2) or siControl were then delivered to mice using the same approach. Following the injections, islets from mice were then harvested and studied by immunoblot, real-time RT-PCR, GSCa, and GSIS. As shown in Fig. 4B, islet Set7/9 protein was reduced by at least 60% by treatment with either siRNA compared with control. Interestingly, similar knockdown was not observed in liver, brain, or muscle, suggesting that for the specific protocol employed, targeting of islets appeared to be most efficient.

Consistent with the data observed in βTC3 cells, depletion of Set7/9 in primary islets resulted in decreases in MafA, Slc2a2, and Ins1/2 (Fig. 5A). Because decreases in these genes would be predicted to result in loss of glucose responsiveness in islets, we tested islet functionality by GSCa. The GSCa assay is a Fura-2AM dye–based assay that measures the intracellular mobilization of Ca²⁺ in response to glucose and is dependent on an intact signaling cascade that begins with the transport of glucose into the β-cell by GLUT2 and ending in the activation of voltage-gated Ca²⁺ channels; intracellular Ca²⁺ mobilization through this cascade is closely linked to insulin secretion (30). As shown in Fig. 5B, reductions in islet Set7/9 protein using either siSet1 or siSet2 resulted in GSCa responses (from 3 to 11 mmol/l glucose) that were only ∼50% those associated with control. These islets also showed diminished GSIS upon raising glucose from 3 to 11 mmol/l (Fig. 5C). Taken together, these results suggest that Set7/9 is necessary for the transcription of a subset of islet genes that are required for the maintenance of normal islet function.

In prior ChIP studies, we demonstrated that Set7/9 occupies the Ins1/2 genes in βTC3 cells (10). This observation, coupled with the finding that the Ins1/2 gene is hypermethylated at H3 Lys4 (10,11), suggested to us that Set7/9 may be directly responsible for the maintenance of this euchromatin histone modification at this gene. We therefore wondered whether the reduction in Ins1/2 gene activity observed upon knockdown of Set7/9 in this study could be secondary to loss of H3-Lys4 methylation. As shown in Fig. 6A, depletion of Set7/9 in βTC3 cells using either of two siRNAs (siSet97 or siSet98) led to 50–80% reductions in dimethylated H3 Lys4 in the Ins1/2 promoter region as determined by ChIP assay but did not affect dimethylated H3 Lys4 in the coding region of the gene. Importantly, we did not observe any changes to monomethylated H3 Lys4 in either the promoter or coding regions of the Ins1/2 gene (Fig. 6B), suggesting that Set7/9 is associated with a very specific dimethylation effect.

To ascertain the specificity of Set7/9, we next examined the genes Slc2a2 (whose transcription is reduced by loss of Set7/9) and Pdx1 (whose transcription is unaffected by loss of Set7/9). As shown in Fig. 6C and D, we observed a specific loss of dimethylated H3 Lys4 at the proximal Slc2a2 promoter but not at the proximal Pdx1 promoter. Similar results were obtained in primary mouse islets depleted of Set7/9 using siSet1, although the decreases in dimethylated H3 Lys4 at the Ins1/2 and Slc2a2 genes did not quite approach statistical significance (Fig. 7A). These results are consistent with gene-specific effects of Set7/9.

To determine whether Set7/9 could be modifying histones directly at the Slc2a2 gene, we next performed ChIP assays in βTC3 cells to assess recruitment of Set7/9 to this gene. Figure 7B shows that Set7/9 occupies the Slc2a2 gene within its proximal control region (−738 to −523 bp relative to the transcriptional start site) but was not detected at the proximal Pdx1 promoter. These results suggest that the chromatin effects observed at the Slc2a2 gene are the result of Set7/9 recruitment to its promoter.

Pdx1, like Set7/9, is required for the full activation of the Ins1/2 gene (23,31,32). This effect of Pdx1 is apparently related not to the recruitment of RNA polymerase II to the Ins1/2 gene but, rather, to defects in the conversion of RNA polymerase II from its initiation to elongation isoforms (11,23). We therefore wondered whether the maintenance of Ins1/2 transcription by Set7/9 might be related to Pdx1. As shown in Fig. 8A, knockdown of Set7/9 in βTC3 cells using siSet97 led to no changes in Pdx1 occupancy at the Ins1/2 gene. However, we observed that RNA polymerase II recruitment to the promoter region of Ins1/2 (and Slc2a2) was diminished by ∼50% (Fig. 8B). These results suggest that Set7/9 action at the Ins1/2 gene may be distinct from that of Pdx1 and that either Set7/9 itself or its associated dimethylated H3-Lys4 modification is linked to RNA polymerase II recruitment.

The islet β-cell derives many of its characteristics, particularly stimulus-secretion coupling, from the transcription of a unique subset of genes that is controlled by a group of cell-specific transcription factors and cofactors. In this report, we identify a new member of this cohort of islet-specific factors, Set7/9, and demonstrate its role in the maintenance of gene transcription and chromatin structure in β-cell lines and primary islets. Set7/9 was originally identified as a histone H3-Lys4–specific methyltransferase (12,13). Subsequently, Set7/9 was shown to methylate Lys residues in a variety of proteins including TAF10 (18), TAF7 (17), p53 (14–16), the estrogen receptor (19), and possibly Pdx1 (11). To date, however, its role in the maintenance of histone H3–Lys4 methylation at islet genes has only been implied (11) and not directly studied.

We show that Set7/9 is distributed in a limited number of tissues in the mouse but includes those derived from all three major germ layers, including ectoderm (brain), mesoderm (muscle), and endoderm (pancreas). Within the pancreas of both the mouse and human, Set7/9 exhibits a striking islet-specific distribution pattern. The enrichment of Set7/9 in β-cells may occur at the level of transcription, as our studies identified a potential conserved β-cell–specific enhancer located in the distal Setd7 gene that is responsive to Pdx1. Its strongly nuclear pattern of distribution, particularly in β-cells, suggested to us a role for Set7/9 in β-cell gene regulation.

To identify genes regulated by Set7/9 in β-cells, we engaged RNA interference approaches to knock down Set7/9 in insulinoma-derived β-cells (βTC3) and primary mouse islets. Successful knockdown using multiple different siRNAs demonstrated reductions in key genes that mediate stimulus-secretion coupling, including Slc2a2, MafA, and Ins1/2. Consistent with these reductions, we observed impaired GSCa and GSIS in islets depleted of Set7/9. We note that our knockdown studies were of short duration by design in order to obtain information on the most immediate and potentially direct genetic targets of Set7/9. Thus, we cannot rule out the possibility that other genes may contribute to the dysfunctional islet phenotype observed or that more prolonged knockdown may not have broader effects on islet function, replication, or survival. Importantly, effects on islet survival are possible, given that recent studies suggest a role for Set7/9 in the maintenance of inflammatory genes in macrophages (33).

As noted earlier, Set7/9 exhibits specific methyltransferase activity toward histone H3 Lys4. Some studies suggest that it may function as a dimethyltransferase (341), whereas others suggest that it functions exclusively as a monomethyltransferase in vitro (35). Although our data are compatible with the role of Set7/9 as a dimethyltransferase in cells, they do not rule out the possibility that it may indeed function as a monomethyltransferase. For example, it is possible that Set7/9 is very closely linked to a second methyltransferase that completes the dimethylation reaction in cells, such that transient alterations in monomethylation are not detectable in our system. Alternatively, it is possible that catalytic activities observed in vitro are modified in the cellular milieu as a result of the complexity of protein-protein interactions. In this context, recent studies of the H3-Lys9 methyltransferase G9a show that it exhibits differing stoichiometries of methylation in vivo depending on binding to accessory proteins (36).

In the absence of endogenous nuclear localization and DNA binding domains, we presume that Set7/9 must be chaperoned to both the nucleus and target genes through an interaction with transcription factors. We have previously demonstrated that Set7/9 forms an immunoprecipitable complex with Pdx1 (11). In those studies, we also showed that depletion of Pdx1 in βTC3 cells leads to diminution of H3-Lys4 dimethylation at the Ins1/2 gene—an observation that our present studies suggest could be caused by disruption of the Pdx1-Set7/9 complex. Interestingly, however, whereas the depletion of Pdx1 led to defective conversion of RNA polymerase II to its elongation isoform (and not its recruitment to the promoter [11,23]), we observed here that depletion of Set7/9 causes impairments in RNA polymerase II recruitment to the promoter. Our findings are therefore consistent with studies suggesting that Set7/9 stabilizes the transcriptional preinitiation complex by methylation of TAF10 (18).

Taken together, our results suggest a model for Set7/9 action at target β-cell genes. We propose that Set7/9 is recruited to specific β-cell genes (e.g., Ins1/2 and Glut2) through interaction with factors such as Pdx1. This recruitment subsequently leads to dimethylation of H3 Lys4 within the promoter region (either directly by Set7/9 or through interaction with another methyltransferase) and recruitment of RNA polymerase II. The subsequent conversion of RNA polymerase II to its elongation isoform requires other factors that are dependent on Pdx1. The preponderance of studies in the literature suggest that dimethylation of H3 Lys4 appears to be crucial for laying a euchromatic template at target genes, thereby leaving genes poised for transcription. Although we have demonstrated an important role for Set7/9 in the transcriptional regulation of key β-cell genes in mature cells, the role of Set7/9 in establishing transcriptional patterns during the embryonic development of β-cells remains to be determined. In future studies, we will directly address its role in embryonic development using conditional Set7/9 knockout mice.

This work is supported by a grant from the Juvenile Diabetes Research Foundation (to R.G.M.) and by grants R01 DK60581 (to R.G.M.) and T32 GM007055 (to T.G.D.) from the National Institutes of Health.

No potential conflicts of interest relevant to this article were reported.

We thank Dr. C. Nunemaker for use of the Ca²⁺ imaging equipment and Dr. K. Brayman for provision of human pancreas sections.